Interaction information for causal inference: The case of directed triangle
By AmirEmad Ghassami, Negar Kiyavash. 2017
27 citations recorded by the source corpus.
Interaction information is one of the multivariate generalizations of mutual information, which expresses the amount of information shared among a set of variables, beyond the information shared in any proper subset of those variables. Unlike (conditional) mutual information, which is always non-negative, interaction information can be negative. We utilize this property to find the direction of causal influences among variables in a triangle topology under some mild assumptions.
Subfields: Artificial Intelligence, Computational Theory and Mathematics, Molecular Biology
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Negar Kiyavash
- AmirEmad Ghassami
- Saber Salehkaleybar
- Kun Zhang
- Bernhard Schölkopf
- Dominik Janzing
- David Lopez-Paz
- Jan Peters
In Computational Theory and Mathematics
- Negar Kiyavash
- Saber Salehkaleybar
- AmirEmad Ghassami
- Kun Zhang
- Bernhard Schölkopf
- Biwei Huang
- Clark Glymour
- Jakob Runge
In Molecular Biology
- Michael Lauber
- Sandra Reitmeier
- Markus List
- Tim Kacprowski
- D. Haller
- Kun Zhang
- Jan Baumbach
- M. Matchado
Citations in the corpus
- Learning Causal Structures Using Regression Invariance (2017, background): “Note that all the triangle DAGs are in the same Markov equivalence class and hence, using the information of one environment alone, observation only setting cannot lead to identification, which makes this structure challenging to deal with [8].”
- Network analysis methods for studying microbial communities: A mini review (2021, methodology): “It employs four different approaches, such as Sign Pattern [69], Overlap [68], Interaction Information [69,70], and Data Processing Inequality [71,72] to identify indirect (environmentally-driven) edges.”